Reflections and Recommendations on Regulating AI Impersonation of Public Figures in Livestream Marketing

Open a certain livestreaming platform, and a well-known actress is enthusiastically promoting skincare products—yet at the very same moment, another “her” is hawking agricultural products in the next livestream room. Since 2025, similar scenes have become increasingly common: an “Olympic champion” selling free-range eggs, a “CCTV anchor” advertising fish oil supplements, and even a “doctor from a top-tier hospital” appearing in a wellness livestream to peddle weight-loss meal plans. These schemes use AI-generated video to replicate the facial features and voices of public figures for product promotion, making it extremely hard for the public to tell real from fake, misleading vast numbers of consumers and seriously harming the public figures concerned. As lawyers specialising in intellectual property and online reputation management, the authors have recently received several inquiries from well-known individuals who complain that these AI-generated fake videos associate them with inferior products, tarnish their reputations, and even generate negative publicity.

The proliferation of such AI impersonation marketing has already attracted serious attention from regulators. Recently, the cyberspace authorities launched a concentrated clean-up campaign against this conduct. As of November 2025, authorities nationwide had dealt with 11,000 AI livestream accounts impersonating public figures and removed over 8,700 pieces of violating marketing content, involving goods ranging from food and cosmetics to health supplements and many other categories. Behind AI impersonation marketing, however, lies a fully developed industrial chain. The technologies and tools it needs can easily be obtained online at low cost. For example, a basic annual VIP package on a certain app for generating face-swap videos costs only 499 yuan; video face-swap services sold on second-hand trading platforms cost as little as 9.9 yuan per instance; and face-swap tool installation packages can be purchased for just a few yuan. Increasingly rampant AI impersonation not only crosses the bottom line of personality rights protection but also breaches the red lines of market competition and advertising regulation, infringes consumer rights, and poses new challenges to reputation governance and legal regulation.

I. The Damage of AI Impersonation of Public Figures to Personality Rights and Market Order

The illegality of AI impersonation marketing lies first in its direct infringement of public figures’ personality rights, including portrait rights and voice rights. Article 1019 of the Civil Code explicitly prohibits the use of information technology to forge another person’s portrait, and Article 1023 provides that the protection of a natural person’s voice shall apply the relevant provisions on portrait rights protection mutatis mutandis. The judgment in case (2023) Jing 0491 Min Chu No. 121XX by the Beijing Internet Court clearly held that the scope of voice rights protection encompasses AI-synthesised voices. Where the general public or a certain segment of the public can identify a specific natural person based on timbre, intonation, and pronunciation style from a voice processed by AI technology, that voice falls within the protection scope of the natural person’s voice rights. In livestream marketing, the unauthorised use of AI-synthesised portraits and voices of public figures falls within the Civil Code’s protection scope for portraits and voices, infringing upon personality interests. In addition, if a public figure’s name is used in the marketing, the right to one’s name may also be infringed; false commercial associations may lower the public figure’s standing in society, potentially infringing the right of reputation; and the unlawful collection and use of personal information such as facial data and voiceprints in the AI face-swap process may also infringe the public figure’s rights in personal information.

Beyond infringing personality rights, such conduct often also constitutes unfair competition and false advertising. AI impersonation marketing fabricates endorsement facts by forging public figures’ portraits and voices, which not only easily causes consumer confusion but also constitutes false representation in product promotion, violating Article 7 (prohibiting confusion) and Article 9 (prohibiting false statements) of the Anti-Unfair Competition Law. Meanwhile, Article 4 of the Advertising Law provides that advertisements shall not contain false or misleading content, and Article 33 expressly requires prior written consent for the use of another person’s name or image in advertising. The unauthorised use of public figures’ portraits and voices in AI impersonation marketing not only violates the principle of truthfulness in advertising but also contravenes the prior-consent requirement of the advertising law, and is a typical case of false advertising.

The Measures for Labeling AI-Generated Synthetic Content, which took effect in September this year, are an important tool for regulating this type of conduct. The Measures expressly require that AI-generated content must bear both explicit and implicit labels, and that providers of online information content dissemination services such as short-video and livestreaming platforms must verify and prominently notify users of relevant content. In practice, however, there are still many ways of evading this obligation that regulators and platforms need to keep watching: some generate content with overseas tools or privately built systems to escape platform supervision; others use mirror-flipping, picture-in-picture masking, and other techniques to get around the labelling requirements. When AI-generated content spreads across platforms, the infringing links become hard to trace, making it difficult to hold technology providers accountable. In addition, most platforms lack the technical capability to identify AI content proactively and often remove violating content only after complaints or reports, so enforcement keeps running into repeat violations, with banned infringers simply switching accounts and carrying on.

II. Building a Multi-Stakeholder Collaborative Regulatory System

Precisely because AI impersonation violates numerous legal provisions, multiple legal tools are available for regulation. Faced with the complex landscape of AI impersonation marketing, a governance system involving regulators, platforms, public figures, consumers, and other stakeholders should be built, combining the forces of technology, law, and public opinion.

From a regulatory perspective, it is recommended that regulators step up oversight of such conduct, hold enterprises to their primary responsibility, and promote cross-departmental collaboration by establishing an AI infringement blacklist. Impersonating merchants investigated by market regulation authorities and violating accounts dealt with by cyberspace authorities should all be recorded in a unified national database, so that the governance effect of “one violation, nationwide ban” is achieved.

As the channel through which AI impersonation content spreads, platforms need to assume the responsibility of the first line of defence, shifting from passive response to proactive prevention. Technically, platforms should deploy AI identification tools and upgrade them in good time to strengthen detection of AI-generated content. On the management side, platforms should offer mechanisms that make it easier to verify and manage public figures’ authorisations. YouTube’s recently adopted likeness detection technology offers a useful reference: public figures can submit their identity documents and a short selfie video in advance; once the application is approved, the system automatically detects and notifies them of all videos featuring their likeness. Public figures can then choose, depending on the situation: 1) if they believe the likeness is AI-generated, submit a likeness removal request; 2) if they believe the video uses copyrighted content, submit a copyright removal request; 3) if the video does not involve an AI-manipulated likeness or is unrelated to them, choose to archive it.

Public figures should first consider using current digital evidence-preservation technologies to secure records of their digital image and voice, laying a solid foundation for later enforcement of their rights. For example, it was reported in April this year that China’s first batch of intellectual property registration certificates for AI digital human data had been issued. Once AI impersonation by others is discovered, they should take active steps to protect their rights, paying attention to prompt evidence preservation and reporting through multiple channels. When preserving evidence, forensically sound collection tools can be used to timestamp the infringing livestream footage, product links, transaction records, and so on, ensuring the evidence chain is complete. After collecting evidence, they should immediately notify the platform to remove the infringing content, file complaints with the cyberspace and market regulation authorities at the same time, and bring a civil action in the people’s court.

Ordinary consumers, as the direct audience of AI impersonation, also need to enhance their awareness and ability to identify AI-impersonated content. AI-impersonated content can be spotted by three details: first, check whether there is a prominent label as required by the Measures for Labeling AI-Generated Synthetic Content—if a livestream room or video is labelled only “virtual anchor” without saying whether it is AI-generated, be on guard; second, watch the interaction and expressions—AI digital humans often respond slowly to unexpected questions, such as “how do you say this in your hometown dialect” or “what are your recent works”—and if the anchor dodges, gives irrelevant answers, or looks unnatural, the content is most likely AI-synthesised; third, verify where the products come from—public figures usually announce collaborations through their official accounts, so if the products recommended in a livestream room are not mentioned through official channels, be cautious before ordering. In addition, once deceived, consumers should promptly save order screenshots and livestream recordings and claim “refund plus triple compensation” under Article 55 of the Consumer Rights Protection Law, which both protects their own rights and gives regulators leads for investigation.

III. Letting AI Technological Innovation Advance on the Track of the Rule of Law

Curbing AI impersonation marketing does not mean throwing out technological innovation with the bathwater, nor does it mean letting things drift while rights are infringed. Only with clear legal rules to draw the boundaries, advanced technological means to strengthen prevention and control, and diverse collaborative mechanisms to pool strengths can AI technology return to its purpose of doing good and marketing return to its foundation of honesty and integrity. This protects not only the rights of public figures but also market order and consumer trust, and is essential to the high-quality development of the digital economy.

Special Disclaimer:

Beijing Dacheng Law Offices, LLP (Shanghai) strictly observes its obligation to protect client information. All client project content referenced in this article is derived from publicly available information or has been obtained with client consent. The content and views expressed herein are for reference only, do not represent any position of Beijing Dacheng Law Offices, LLP (Shanghai), and shall not be construed as providing any form of legal opinion or advice. For reprinting or citation of any content from this article, please contact us via private message regarding authorisation, and indicate the source at the beginning of the article when reprinting. Without authorisation, no content from this article may be reprinted or used.